AI Voice Receptionist SaaS (Next.js + OpenAI)

The Core Problem: Clinics, salons, and gyms lose revenue daily to missed calls. Existing AI bots couldn't be trusted to actually book appointments accurately without hallucinating, and business owners had zero visibility into how the AI made decisions.

The Technical Solution: I architected a multi-tenant SaaS platform from scratch using Next.js and MongoDB.

• AI Architecture: Built an onboarding wizard with a prompt compiler and a robust tool-calling engine.

• Safety Guards: Implemented strict server-side booking guards so the AI can only propose a booking, while the server validates it.

• Transparency: Developed a 'Flow Studio' node graph that visually exposes the AI's routing logic for the business owners.

Measurable Business Impact: Delivered a fully functional platform where business owners can visually script, test, and trust AI voice receptionists. The server-side guards ensure safe bookings without hallucination risks, designed to capture 100% of after-hours leads.

Tech Stack: Next.js, TypeScript, MongoDB, OpenAI, Tailwind CSS.